AI Show 237: GPT-6 Astra, Data Centers, and AI in Schools

Sep 8, 2026 · The Artificial Intelligence Show
🎧 PodShort 33 min squeezed to 3 SaaS HoarderSH New
Episode artwork
Paul Roetzer
Founder & CEO at SmarterX and Marketing AI Institute
Mike Kaput
Chief Content Officer at SmarterX
The Artificial Intelligence Show
33 min squeezed to 3
Full episode from The Artificial Intelligence Show
Quotable Moments

We've all known that they knew they were stealing it. Their internal communications indicated that they knew they were stealing it. Every lab knew the other lab was doing it, so they were going to do it. That is not debatable.

Paul RoetzerFounder & CEO at SmarterX and Marketing AI Institute
00:00

The model is dangerous. The model can do things that aren't safe out of the box. And they have spent a bunch of time trying to put things in place to prevent it from doing unsafe things.

Paul RoetzerFounder & CEO at SmarterX and Marketing AI Institute
08:00

I'm not saying it's all naivete. I think they know often what they're doing, but it is wild to see this because I don't read this as like, cool, this can use spreadsheets. I read this as, cool, we don't need people that read spreadsheets.

Mike KaputChief Content Officer at SmarterX
19:00
Key Insights
  • OpenAI's GPT-6 Astra is positioned as a new generation of intelligence, capable of performing complex professional tasks and automating tedious work like filling out forms, updating CRM records, and organizing calendars.
  • GPT-6 Astra demonstrates significant advancements in computer use, achieving a 72.6% score on the OS World 2.0 benchmark and completing tasks 47% faster in simulations, indicating its ability to operate across applications more efficiently.
  • Astra's ability to solve unfamiliar problems is highly advanced, scoring 99.9% on the Arc AGI 3 evaluation, which tests AI's capacity to experiment, learn, and strategize without prior rule explanations.
  • A new technique called recurrent depth or a looped transformer, used in Astra, allows the AI to improve answers by processing text multiple times but also obscures its reasoning, making it harder to monitor for bad behavior.
  • The increased capability of AI models like Astra, particularly in computer use, is leading to a situation where the technology can perform tasks previously done by humans, raising concerns about job displacement and the need for responsible adoption.
  • The public and political response to AI is becoming increasingly polarized, with some advocating for bans and others pushing for rapid development, creating a challenging environment for responsible AI integration.
  • The U.S. Justice Department's stance that training AI models on copyrighted material constitutes fair use, while model output might raise different copyright questions, is a critical development in ongoing legal battles.
  • Major financial institutions are pressuring law firms to lower legal bills or change their charging models due to AI's ability to speed up routine legal work, signaling a shift towards value-based pricing in professional services.
Metrics Mentioned
  • 72.6% (GPT-6 Astra's score on the OS World 2.0 benchmark, indicating its ability to carry out workflows across applications.)
  • 47% (Reduction in time per task for GPT-6 Astra in simulations compared to previous models.)
  • 99.9% (GPT-6 Astra's score on the Arc AGI 3 evaluation, demonstrating its ability to solve unfamiliar problems.)
  • 97.6% (GPT-6 Astra's score on the hardest tier of Frontier Math, up from 83%.)
  • $1 billion (OpenAI's ChatGPT ads' annualized revenue run rate, achieved less than 200 days after launch.)
🎧Full Episode:The Artificial Intelligence Show →